分子模型的构建

1、 分子模型的构建（1）Molecular Modelling Experiments (1)Institute of Applied and Metallurgical Physical ChemistryCentral South UniversityModel Building 1 Introduction1.1 Theneed for computer modellingTraditionally chemists have synthesized molecules and then tested their properties by experiment. In the pharm

2、aceutical industry , for example, it was in the past common for a company to produce hundreds of compounds for every one worth pursuing into the clinic. This is naturally very expensive. If instead a compound with the desired properties can be designed merely by executing an appropriate computer pro

3、gram, there are obvious benefits both intellectually and commercially. The same situation is in new function materials field.This hope has become in large part reality. All major pharmaceutical and agrochemical companies now employ specialists in computer-aided molecular design and the synthetic che

4、mists are increasingly using these facilities for themselves. Long before the advent of computers, chemists used models to aid their understanding of molecules. Basically these were of two types: wire models such as the Dreiding sort which represent bonds by fine metal tubes giving an idea of the mo

5、lecular skeleton, and space-filling models like the CPK (Corey, Pauling, Kolthun) type which represent each atom by a sphere of the appropriate radius and give an idea of the electronic “flesh”.These types of representation have been carried over into the computer graphic molecular modelling program

6、s, as you will see during this experiment. There are, however, major advantages in the computer displays over mechanical models. Molecular geometries can be drawn from crystallographic databases so that they are realistic rather than average values of bond lengths and angles; the display is created

7、in moments rather than weeks for large molecules; the pictures are not subject to gravity (!); the images may be manipulated interactively and parts may be removed or added. It is possible, for example, to observe the active site of an enzyme from inside the protein. In fact the mostsophisticated mo

8、delling programs rival flight simulators, which is not surprising since the hardware and many aspects of the software have much in common. Most drugs and agrochemicals are small molecules of less than fifty atoms. Almost always the molecules are flexible and evoke their important biological response

9、 by binding to a specific site on a macromolecule, generally a protein or nucleic acid. Although it is possible to learn about the shape or conformation of the small binding partner from X-ray crystallography of the compound, or by nuclear magnetic resonance spectroscopy of solutions, these data are

10、 insufficient to understand the shape of molecules when they bind into the receptor slots of the macromolecule. The small molecules, and indeed the protein, may adapt their shapes on binding. Hence it becomes important to calculate the energy of the drug as a function of its shape so as to be able t

11、o answer the question "into what shapes could this molecule be distorted on binding without expending more than a given energy threshold?" It is this latter question which is answered by energycalculations. The computer performs the calculation and a graphics terminal can display the resul

12、ts.1.2 Calculating molecular propertiesTwo quite distinct approaches are used in the computation of the energy of a molecule as a function of its shape. The purer of the possibilities is the use of molecular quantum mechanics to solve the Schrödinger equation in an approximate Hartree-Fock mann

13、er. These computations may be of the so-called ab initio variety in which all the many millions of integrals implied in the theory are rigorously evaluated. The alternative is a semi-empirical calculation, in which time is saved by replacing computed values of integrals by experimentally determined

14、parameters. Quantum mechanical calculations are the favoured approach if, in addition to a value of the energy or stability of a molecule, we also require some other molecular properties which can be derived from the molecular wavefunction (such as dipole moment, electrostatic potential or energies

15、of individual molecular orbital).A different strategy is to use the so-called molecular mechanics methods. These are purely empirical. The methods treat a molecule as a collection of balls (the atoms) and springs (the chemical bonds). Potentials are introduced for all atom-atom interactions; for the

16、 bending of bond angles and for torsional changes. The number of disposable parameters fixed by forcing agreement with experimental quantities, such as heats of formation, may be as large as a hundred, but over the years these potentials have been refined to the point where molecular calculations ar

17、e reliable for many types of molecule even when they contain heteroatoms like sulphur or rings of atoms.Once the energy is calculated, the range of shapes which a molecule may adopt may be compared for molecules which fit a receptor site of interest with the hope of learning something about the dema

18、nds of the binding process.1.3 Structure-activity relationshipsIn addition to computing the energy of an isolated small molecule, other calculable properties may be of utility, especially in the search for a relationship between molecular structure and biological activity. From the molecular wavefun

19、ction which is, along with molecular energy, the output from a quantum mechanical calculation, a large variety of molecular properties may be calculated. The square of the function at any point in the space surrounding a molecule is, of course, a measure of electron density. We may also compute at t

20、he same point the interaction energy of a unit charge (the electrostatic potential) or its gradient, the electrostatic field, which is a measure of the interaction energy a dipole would have at that position. Computer graphics permit representations of these important quantities to be made using col

21、our coding and stereoscopic views. Such graphical displays permit qualitative connections to be made between properties and activity.For quantitative structure activity relationships (“QSARs”) the numerical values of computed quantities are used in regression analyses to provide statistically suppor

22、ted predictions of the activity of hypothesized new compounds. The parameters used in this way include the charges on specific atoms; the electron density in bonds; the availability of electrons as judged by orbital energies; the capacity to accept electrons and perhaps most effectively the properti

23、es of the so-called frontier orbitals - those electrons which are least tightly bound by the molecule.1.4 Protein-drug calculationsThe use of computers described so far has concentrated on calculating and displaying properties of the small molecules which act biologically by binding to receptor site

24、s on macromolecules. Computational techniques have been developed to the extent that the more logical approach of including the protein or nucleic acid macromolecule is now also possible. This adds a whole new dimension to molecular design.The first requirement is knowledge of the molecular structur

25、e of the large molecule binding partner. In ideal circumstances this comes from X-ray crystallographic studies of crystals of the macromolecule. High resolution refined crystal structures are now available for large numbers of proteins, mostly enzymes and for some oligonucleotides. Based on these cr

26、ystal structures the architecture of similar proteins can be derived by a combination of computer techniques including recognizing similar structural elements, theoretical calculation of relative energies and computer-aided model building. This type of work is becoming increasingly important as gene

27、 sequences (and hence the amino acid secondary structure of proteins) become more readily available. Even artificial intelligence techniques are being used to predict the structures of the binding sites of receptors.If we have a reasonable knowledge of both the macromolecular binding site and the sm

28、all molecular partner then the calculation of their binding energy is a most important quantity which can come from theoretical computation. If we want to design an inhibitor of an enzyme then that inhibitor must bind to the active site much more tightly than the natural substrate. If that inhibitor

29、 is to act as a drug then the stronger it binds, the smaller the dose which will be required medicinally.The actual calculation of binding energy can be done using the same types of computer program which are used to calculate the energetic properties of isolated small molecules. In quantum mechanic

30、al calculations, inclusion of every atom of receptor protein is too demanding in terms of computer time, so approximations such as treating atoms as point partial charges are introduced. The molecular mechanics calculations by contrast are rapid enough to cope with the tens of thousands of enzyme at

31、oms. Both approaches have now reached a level of sophistication where the calculations of binding energy are in good agreement with experimental binding enthalpies so that novel inhibitors can now be designed.1.5 Introductionof molecular motionThus far molecular design has been treated almost like e

32、ngineering design: molecules must fit together and bind tightly but they have been considered to have definite static shapes. In reality, of course, molecules are in dynamic motion. When a small molecule binds to an enzyme, it is not a key fitting a lock but a throbbing vibrating pair of species whi

33、ch interact. These dynamic effects need no longer be ignored. Since the molecular mechanics potentials can be computed readily, the gradients of the potential and dynamic behaviour may be predicted usingNewton's laws of motion. This use of so-called molecular dynamics has become particularly imp

34、ortant now that experimental dynamic properties are emerging from nuclear magnetic resonance studies. It has added an extra level of realism to computer-aided molecular design and again is much clarified by using computer graphic displays to illustrate the motions involved.2 ProcedureThis experiment

35、 is run on a PC in the lower teaching laboratory. No special knowledge of computing is required.The experiment is in two parts. In the first, you will use the molecular modelling package Gaussian98(Chemoffice5.0 or Hyperchem5or C2(Castep, Demol3) to complete a number of straightforward tasks, whose

36、aim is to allow you to become familiar with the software. In the second, you will devise an exercise which uses Spartan to solve some appropriate chemical problem. You may find it easiest to propose a suitable project after you have spent some time becoming familiar with the program and have an clea

37、r understanding of what it can do, but it is essential that you discuss your ideas with a demonstrator before you start work on any project.The project forms the bulk of this experiment, and may require several hours work with the modelling program and possibly some time spent investigating the rele

38、vant literature. Once your exercise is approved, you will:(i) use the molecular modelling package to investigate your topic;(ii) prepare a suitable write-up, discussing your results, comparing them to any found in the literature and considering their reliability.A copy of your project will be added

39、to the file beside the experiment, so that later students can learn from the approach you have taken, and any extra features of Spartan which you have used but are not described in these notes.3 StartingA ProgramGaussian98 (Chemoffice5.0 or Hyperchem5 or C2(Castep, Demol3)is a powerful program which

40、 can be run on a PC or Unix workstation. It requires the presence of a hardware key (a “dongle”) connected to the rear of the computer. If the program will not run, check with the technician that the dongle is present. To start a program, double click on the program icon. The principle aim of this e

41、xperiment is to use the program to solve a real chemical problem. In order to do this, you will need to become familiar with various aspects of the program, and the best way to do this is through the completion of tasks which illustrate the capabilities of the software. Print out or record in your d

42、ata book any graphics or data which seem interesting, but do not print everything, otherwise you will collect huge volumes of paper!Task 1. Building and manipulating a small moleculeTo get you started, you will first build and display a small molecule, such as HF/ H2O/ CH4/ C2H3Br /C6H6 /(CBrH2-O-CH

43、3). et al, Once the molecule is built, you will determine its equilibrium geometry and dipole moment. New molecules are built by selecting atomic fragments from a menu and joining them as required. As you will discover later, it is possible to specify geometry, stereochemistry, charge, multiplicity

44、and other characteristics of the molecule.a) Click the left hand mouse button (lmb) on the Filemenu and select New. (Alternatively, you can click on the New icon, which is the leftmost icon.) A panel of common molecular fragments will appear, with tetrahedral carbon already highlighted. Move the cur

45、sor to the middle of the green display area and click the lmb to place a tetrahedral carbon as the starting point for your structure.b) Click on the -Brbox in the fragment panel and then click the lmb over the end of one of the carbon single bonds to add the bromine.c) Select from the fragment panel

46、 the singly-bonded oxygen atom and add that to the structure.d) Finally add a second tetrahedral carbon to the free oxygen bond. All remaining free valences will automatically be completed with hydrogen atoms.You can rotate the molecule by pressing and holding down the lmb in the display area and mo

47、ving the mouse. You can move the molecule using the rmb. Try both now.e) You can determine the molecular mechanics strain energy by selecting Minimise from the Buildmenu. (Alternatively, you can use the icon for this, which is the E with a downward-pointing arrow). The geometry will adjust rapidly a

48、s the calculation is in the progress, and the final strain energy will be displayed at the bottom right, together with the molecular point group. Record both of these in your data book.The molecule can be displayed on screen in several different formats.f) Choose View from the Build menu and note th

49、e effect of selecting different types of display from the Model menu. The molecule can be rotated and moved in each display type. Print out one of the structures if you wish.g) Return to the ball and wire model. In this format (and in the wire model) it is possible to display atom and bond labels. S

50、elect Configure Labels.from the Modelmenu and make sure that Labelsin the Modelmenu is ticked. Click OK, and the select Labelsfrom the Modelmenu. You can remove labels by again selecting Labelsfrom the Modelmenu.Task 2. Determining molecular geometryIt is simple to make measurements of the geometric

51、 parameters for your molecule.a) Return to the ball-and-stick model. Select Measure Distancefrom the Geometrymenu. Click the mouse on two atoms in succession, or (if you wish to measure the distance between bonded atoms) on the bond which joins them to display the distance between the atoms. Measure

52、 the length of each of the C-O bonds and record the distances in your data book. Comment on the values.b) Determine the C-Br bond length and compare this to the average value quoted in any standard chemistry textbook.c) Choose Measure Anglefrom the Geometrymenu. Click three atoms in succession (or o

53、n two bonds sharing a common atom) to determine the angle between the two bonds. Determine the angles around the bromine-bonded carbon atom and comment on their values.Task 3. MO CalculationsA variety of parameters can be found through a MO calculation. The first step required is to choose the type

54、of calculation.a) Select Calculations.from the Setupmenu. In the dialogue box which opens up select Equilibrium geometryfrom the leftmost pull-down menu. Select Hartree-Fock and 3-21Gfrom the two rightmost menus to specify a Hartree-Fock calculation using the 3-21G split-valence basis set. Verify th

55、at the Total Chargeis Neutraland the Multiplicityis singlet. Click on OKto remove the dialogue.b) Select Submitfrom the Setupmenu. A file browser appears. Choose the default name (probably "Spartan1" or something similar) and click on Save. This step will start your calculation off-line. C

56、lick OKto remove the message.c) Once the calculation has completed (between ten seconds and several minutes will be requireddepending upon the complexity of the calculation and speed of the machine) you will be notified. You can, if you wish, inspect output from the job using Outputin the Displaymen

57、u. Of particular interest is the dipole moment. To find this, select Propertiesfrom the Displaymenu. Record the dipole moment and compare it to the literature value for this molecule of 2.06 Debye(CBrH2-O-CH3).Task 4. Displaying electron density and potential energy surfacesa) Select Surfacesfrom th

58、e Setupmenu. Click on Add. Now select densityfrom the Surfacemenu and nonefrom the Propertymenu. Click a second time on Add.and this time select densityfrom the Surfacesmenu and potentialfrom the Propertymenu. This specifies the calculation of an electron density surface onto which the value of the

59、electrostatic potential has been mapped. Click on OK.b) Submit the job (Submitfrom the Setupmenu); you do not need to remove the dialogue window to do this. When the calculation is complete again select Surfaces. Click on the line density Completed 0.002. Click on the density surface. A menu will appear at the bottom right of the screen. Note the effect of changing the display style and print out one of the figures.Select Closefrom the Filemenu to remove the molecule from the scr